How to find the unknown coordinates reference system for my map data

Kofi_Lartey · ‎03-21-2016

I have several shapesfiles which perfectly overlay but the coordinates reference system is unknown. I need to overly these with other data in a differetn coordinate system. The coordinate extent is quite unique. I tried initially to overlay with several projected and geographic coordinates systems associated with the area out of which two showed very close displacement (about 60m). Next i tried different datum with one of the 'promising' coordinates reference systems (a geographic coordinates system) and also change the units but same results. Attached is a clip of two of the coordinate extents. I am yet to try the second 'promising' CRS but decided to put it out just in case i am missing something

Kofi_Lartey · ‎03-25-2016

Rebecca, I am trying to project from the current defined CRS (Leigon_Transverse_mercator) to a new one (Ghana meter grid). There are only slight differences between the two, both uses the same Geographic coordinates system (GCS_ Leigon), projection (transverse mercator), datum (D_Leigon), but different units; one is in meters and the other mm, central meridian and latitude of origin are different. All others the same ie false northern and eastern, scale factor. So far the transformation comes with the error message 'transformation is invalid'. Any ideas where i need to improve?

MelitaKennedy · ‎03-25-2016

Hi Kofi,

That message occurs when either the input or output geographic coordinate systems don't match the internal definitions. So the software sees them as "different" so thinks that there should be a geographic/datum transformation OR there's no transformations available between two "known" GCS.

A workaround if the input/output GCS are supposed to be the same is to add the data to ArcMap and set the data frame to the coordinate system you want to convert to.

Ignore any geographic/datum transformation warning message.

Right-click the data layer in the table of contents and choose data, export data. In the dialog, set it to use the data frame's coordinate system and export the data.

Melita

NeilAyres · ‎03-26-2016

As Melita says, if both underlying GCS / datums are Leigon, then no transformation is necessary.

The WKID for Ghana_Metre_Grid is 25000.

EPSG has this definition, the units are metres (obviously). Why are you talking about mm?

And, quite what is the definition of your starting coord sys. Also the older "Accra" system was in Gold Coast feet.

Kofi_Lartey · ‎03-26-2016

it was based on this that i developed the projection coordinates system and as you can see the units is 0.001 which when i checked was mm. I have tried Melita's suggestion, but when i tried to open or load in arcmap nothing showed up even when i zoomed to layer or the full extent. The attribute table however is fully populated.

DanPatterson_Retired · ‎03-26-2016

You use a comma as a decimal separator.

the xmin to xmax is -0.5 to 2455.5 whatevers

the ymin to ymax is -0.5 to 3576.5 whatevers

the whatevers aren't millimeters... they aren't degrees, unless the study area is small ( is 2.455 ish km wide) then the units aren't meters...

perhaps 0.001 represents a fraction of a decimal degree?

this leads me to question again,.... are thes values representative of the number of cells in the x and y directions.

Kofi_Lartey · ‎03-27-2016

Dan, i get your drift, but this is a vector file so how do i get the number of cells in the x,y direction. Or maybe i should convert to raster? I called the source of the data and he was of very little help except to suggest that its possible the document has been corrupted.

DanPatterson_Retired · ‎03-27-2016

It looks like the file you have might have been converted to vector from raster. I suppose that has been ruled out already

curtvprice · ‎03-27-2016

Kofi Lartey that extent really looks like it the vector data came from an unprojected raster dataset of size 2456 rows x 3577 columns. If you have no idea where this data came from and can identify some tie points, you may want to try the Transformation toolbar to warp your data to match some other dataset.

MelitaKennedy · ‎03-28-2016

I had to go track down an ArcInfo workstation book to look up the "Units 0.001" definition. A custom unit's value is defined as units per metre. So that stands for kilometres, not millimetres. Even with that information, the coordinate extents still don't make sense unless the data covers Ghana plus a bunch of countries to the north and east.

If I very roughly check Ghana's extents against the 2455x3576 "cell sizes", I get around 5 meters per "cell". That's not too bad for a raster that might have represented soil or rainfall data.

I don't know if you're willing to send one of the files (shapefile, .e00, whatever) to me (mkennedy at esri dot com) so I can look at it.

Melita

MelitaKennedy · ‎03-29-2016

Okay, thanks to Kofi Lartey, I got the e00 files. One, a country boundary file with at least level 1 administrative districts, is in a custom projected coordinate system, complete with a custom ellipsoid.

Projection TRANSVERSE

Datum USER_DEFINED 0.000 0.000 0.000 0.0000 0.0000 0.0000 0.0000

Zunits NO

Units FEET

Xshift 0.0000000000

Yshift 0.0000000000

Parameters 6378295.0000000000 6356746.0000000000

0.99975000 /* scale factor at central meridian

-1 0 0.000 /* longitude of central meridian

4 40 0.000 /* latitude of origin

274320.00000 /* false easting (meters)

0.00000 /* false northing (meters)

The projection parameters are the Ghana National Grid / Metric Grid (although the false easting differs by about 0.5 meter). The custom ellipsoid has a flattening around 1/295.9903. The custom ellipsoid is close the "War Office" so there may be a Gold Coast foot/metre conversion issue.

Anyway, the other two coverages are definitely converted from rasters. Here's a close-up:

Also complicating the issue is that these two files don't share the same country boundary. There could be a datum difference between them, but the offsets are not regular, so it's hard to say for certain.